[458] in Pthreads mailing list archive
Re: pthread_cancel points
daemon@ATHENA.MIT.EDU (Bob Knighten)
Mon Oct 7 18:39:01 1996
Date: Mon, 7 Oct 96 15:13:38 PDT
From: Bob Knighten <knighten@co.intel.com>
To: proven@cygnus.com
Cc: streepy@healthcare.com, pthreads@MIT.EDU
In-Reply-To: <199610070719.DAA00198@qed.proven.org>
Reply-To: knighten@co.intel.com (Bob Knighten)
Christopher Provenzano writes:
>
> > It seems really odd to me that the draft 10 standard list of cancel
> > points doesn't include a single network/socket operation.
> >
> > That seems very wrong. Why would you make "read" a cancel point, but
> > not "recv" or "recvmsg". Things like "accept" and "connect" are also
> > excluded.
> >
> > Admittedly, I don't have the whole spec, just certain sections.
> >
> > So, am I missing something, or does the spec specifically ignore socket
> > operations?
>
> The POSIX thread standard POSIX 1003.1c is an extension of the
> POSIX 1003.1 standard, which didn't include any socket or network
> operations. I expect that when the network API is standardized then
> those operations will be cancelable but first the have to standardize
> what they are.
>
That is exactly right, though I would mention that the POSIX thread standard
is actually a part of the revised 1003.1 standard. The actual reference is
ISO/IEC 9945-1(E)
ANSI/IEEE
Standard 1003.1, 1996 Edition
Information technology - Portable Operating System Interface (POSIX) -
Part 1: System Application Program Interface (API) [C Language]
--
Robert L. Knighten | knighten@co.intel.com
Server Architecture Lab |
Intel Corporation |
5200 N.E. Elam Young Parkway, CO3-202 | (503) 677-4315
Hillsboro, OR 97124-6497 | (503) 677-6700 [FAX]
